resin t2 loucl 10 ethernet · the two tasks of the network-access-layer are divided into two...

13
RESIN T2 10. Ethernet Claude Loullingen 1 10. Ethernet 10.1 Network-Access-Layer (TCP-IP-model) Already in chapter 2 we have seen that the network-access-layer builds a SDU so called frame. The header of this frame contains amongst other information the destination and source-MAC-address. The network-access layer has however a second task which is to convert the frame containing bits into a signal that can be transported over a medium. medium signal used copper voltage air or vacuum electromagnetic wave optical fiber light The two tasks of the network-access-layer are divided into two different layers in the OSI- model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1 Data-Link Layer (Layer 2) The data-link-layer builds the frame. It does not only add the MAC addresses to the header but also adds information to detect and may be correct transmission errors (channel coding). 10.2.2 Physical Layer (Layer 1) The physical layer converts the frame into a signal. 10.3 Ethernet standards Ethernet is a family of standard that define both the Layer 2 protocols and the Layer 1 technologies. Ethernet is the most widely used wire based LAN technology and supports transmission speed of 10, 100, 1000, or 10.000 Mbps.

Upload: others

Post on 05-Oct-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 1

10. Ethernet

10.1 Network-Access-Layer (TCP-IP-model)

Already in chapter 2 we have seen that the network-access-layer builds a SDU so called

frame. The header of this frame contains amongst other information the destination and

source-MAC-address.

The network-access layer has however a second task which is to convert the frame

containing bits into a signal that can be transported over a medium.

medium signal used

copper voltage

air or vacuum electromagnetic wave

optical fiber light

The two tasks of the network-access-layer are divided into two different layers in the OSI-

model.

10.2 DataLink- and Physical-Layer (OSI-model)

10.2.1 Data-Link Layer (Layer 2)

The data-link-layer builds the frame. It does not only add the MAC addresses to the header

but also adds information to detect and may be correct transmission errors (channel coding).

10.2.2 Physical Layer (Layer 1)

The physical layer converts the frame into a signal.

10.3 Ethernet standards

Ethernet is a family of standard that define both the Layer 2 protocols and the Layer 1

technologies. Ethernet is the most widely used wire based LAN technology and supports

transmission speed of 10, 100, 1000, or 10.000 Mbps.

Page 2: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 2

Examples of Ethernet standards and characteristics:

standard speed cable Conector maximum

length

10Base-T 10 Mbit/s 4 twisted pair min. cat. 3 RJ-45 100m

100Base-TX 100 Mbit/s 4 twisted pair min. cat. 5 RJ-45 100m

1000Base-T 1000 Mbit/s 4 twisted pair min. cat. 5 RJ-45 100m

10GBase-T 10.000 Mbit/s 4 twisted pair min. cat. 6 RJ-45 100m

10.4 Layer 2 in the Ethernet standards

The Ethernet frame does not only add a header to the packet but also a trailer. The trailer

contains a 32bit FCS (frame check sequence). These 32bit allow detecting with a very high

probability if a transmission error occurred or not. The technique used to calculate the FCS

in all Ethernet standards is the CRC (cyclic redundancy check). See also:

http://en.wikipedia.org/wiki/Computation_of_cyclic_redundancy_checks

The CRC allows only detecting an error but not correcting it.

10.5 Layer 1 in the Ethernet standards

The layer 1 is called the physical layer. This layer defines the means of transmitting raw bits

rather than logical data packets. Some of the points defined in the physical layer are:

• connector, cable and wiring

• voltage levels

• signal type

Page 3: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 3

10.5.3 The Ethernet connectors

RJ45 connectors can be used up to frequencies of 500MHz. This corresponds to category

6a cables.

For frequencies higher than 500 MHz another connector must be used, for example a GG45

connector.

Page 4: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 4

10.5.4 The Ethernet cables

The current Ethernet standards know 3 types of cables. All consists of 4 twisted pairs.

Twisting the pairs is done in order to eliminate crosstalk (deut.: Übersprechen) and any other

type of interference the cable might pick up from nearby electrical sources.

Page 5: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 5

Categories of cables:

Name Typ Bandwidth Application

Cat1 UTP 0,4 MHz Telefon- und Modem-Leitungen

Cat2 UTP 4 MHz Ältere Terminalsysteme, z. B. IBM 3270

Cat3 UTP 16 MHz 10BASE-T and 100BASE-T4 Ethernet

Cat4 UTP 20 MHz 16 Mbit/s Token Ring

Cat5 UTP 100 MHz 100BASE-TX & 1000BASE-T Ethernet

Cat5e UTP 100 MHz 100BASE-TX & 1000BASE-T Ethernet

Cat6 UTP 250 MHz 10GBASE-TEthernet

Cat6a STP 500 MHz 10GBASE-TEthernet

Cat7 STP 600 MHz 10GBASE-TEthernet

Cat7a STP 1000 MHz 10GBASE-TEthernet

Cat7 S/FTP 600 MHz Telefon, CCTV, 1000BASE-TX über dasselbe Kabel. 10GBASE-T Ethernet.

Cat7a S/FTP 1000 MHz Telefon, CATV, 1000BASE-TX über dasselbe Kabel. 10GBASE-TEthernet.

Cat8 S/FTP 1600 MHz – 2000 MHz Telefon, PoE, 40GBASE-TEthernet.

Relation between bandwidth and transmission speed:

Page 6: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 6

10.5.5 Wiring the RJ45 connector

There are two standards for RJ45 connector wiring:

T568A T568B (most commonly used)

Use of crosscables:

Page 7: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 7

10.5.6 Signals in 100Base-TX

In the 100Base-TX standard only 2 of the 4 twisted pairs are used. On a host pin 1 and 2

are used to transmit and pin 3 and 6 are used to receive data.

Exercise 1:

Complete the following drawings with the pin numbers, signals and connections:

Host Switch

1 TXD+ 1

2 TXD- 2

3 RXD+ 3

4 NC 4

5 NC 5

6 RXD- 6

7 NC 7

8 NC 8

GND GND

Host Host

1 TXD+ 1

2 TXD- 2

3 RXD+ 3

4 NC 4

5 NC 5

6 RXD- 6

7 NC 7

8 NC 8

GND GND

Page 8: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 8

Switch Switch

1 1

2 2

3 3

4 4

5 5

6 6

7 7

8 8

GND GND

10.5.7 Signals in 1000Base-T

1000Base-T uses all 4 twisted pairs for bidirectional transmission.

Host Switch

1 BI_DA+ BI_DA+ 1

2 BI_DA- BI_DA- 2

3 BI_DB+ BI_DB+ 3

4 BI_DC+ BI_DC+ 4

5 BI_DC- BI_DC- 5

6 BI_DB- BI_DB- 6

7 BI_DD+ BI_DD+ 7

8 BI_DD- BI_DD- 8

GND GND

Page 9: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 9

10.6 Galvanic separation and Power over Ethernet (POE)

Page 10: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 10

10.7 Converting a series of bits to an electrical signal

The simplest way to convert a bit stream to an electrical signal would be to assign two

voltage levels to the two binary states, for example:

logical state voltage level

0 0V

1 5V

The first problem with this approach is that the signal has a DC part which cannot be

transmitted through the transformers seen in the previous chapter. In addition the DC part

of the signal would be added to the DC part of the Power over Ethernet.

The second problem is that the spectrum of this signal contains harmonics with frequencies

much higher than the cable bandwidth allows. Explanations:

The "worst case" signal that can be imagined in regards of its spectrum, is a continuous

series of a 01 sequence. This signal has the lowest possible periodic time T. At 1GBit/s a

single bit will need 1ns to be transmitted. The periodic time will then be 2ns which correspond

to a frequency of 500MHz. Thus the fundamental contained in our worst case signal has a

fundamental frequency of 500MHz. The next spectral line in a rectangular signal is the third

harmonic having a frequency of 1500MHz. However the bandwidth of a CAT6a cable will

only allow transmitting up to 500MHz.

Page 11: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 11

distortion caused by the low pass behaviour of a CAT6a cable to a 01 series:

t

5V

0V

2ns

∼∼

fC=500MHz

t

5V

0V

2ns

u(t)u(t)f=500MHz f=500MHz

Theoretically we could still retrieve the binary information out of the distorted signal but in

practice noise will lead to unacceptable high error rates.

The third problem is that on Ethernet no clock signal is transmitted with the signal so that

the receiver needs to reconstruct the clock signal out of the data signal. If the data signal

contains long series of zeros or ones, no clock signal can be extracted and the receiver

clock will start to desynchronize from the transmitter clock.

10.7.8 Channel coding

Channel coding is a technique that allows avoiding one or several of the above mentioned

problems. 100Base-TX Ethernet uses for example the 4B5B-code in order to assure that

within every 4 bit sequence at least one level change is assured. To do this a 5th bit is added

to every 4 bit sequence according to the following translation table:

4B 5B 4B 5B

0000 11110 1000 10010

0001 01001 1001 10011

0010 10100 1010 10110

0011 10101 1011 10111

0100 01010 1100 11010

0101 01011 1101 11011

0110 01110 1110 11100

0111 01111 1111 11101

Page 12: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 12

In order to reduce the maximal fundamental frequency of the 4B5B-signal this one will be

coded a second time with the MLT-3 code. This code uses three voltage levels +1V, 0V and

−1V. Only when a logical 1 appears in the input signal then the output signal will change its

state in the following order: 0V, +1V, 0V, -1V, 0V, etc.

Exercise 2:

Draw the 4B5B signal of the frame bit series. Then draw the MLT-3 signal of the 4B5D signal.

Page 13: RESIN T2 LOUCL 10 Ethernet · The two tasks of the network-access-layer are divided into two different layers in the OSI-model. 10.2 DataLink- and Physical-Layer (OSI-model) 10.2.1

RESIN T2 10. Ethernet

Claude Loullingen 13

Exercise 3:

Draw for each of the following codes the worst case signal in regards of its frequency

spectrum for a 100Mbit/s stream. What will be the fundamental frequency of the signal?